analiza aiqbench
This arXiv study tests whether LLM agents acting as audit associates shift their estimates under social pressure from colleagues, an outside reviewer, and a client CFO, across twelve model settings. The findings are mixed: agents yield to some pressures much like human auditors do, but they often ignore the CFO and respond to career incentives in the opposite direction from people, sometimes naming the incentive as an independence threat. For readers building or deploying LLM decision-support tools in regulated workflows, the key takeaway is that model behaviour under pressure is not human-like and must be evaluated per model and per pressure type.
Do Social Influence Pressures Move LLM Agents? Evidence from Delegated Audit Judgment
arXiv:2611.00508v1 Announce Type: cross Abstract: Large language model (LLM) agents are rapidly being integrated into complex professional workflows such as auditing and entrusted with judgments under human review. Human auditors are known to yield to social influence pressures from superiors, peers and clients, and this paper asks whether LLM agents yield to the same pressures and, if so, how they respond. In an experiment on twelve model settings, an LLM agent acting as a senior associate in audit analytical review forms an estimate and is then challenged without new evidence by colleagues inside the firm, by a reviewer from another firm and by the client's CFO, with and without a career inducement. Experimental findings show that LLM agents yield to pressure from the engagement partner, from peers and from a reviewer from another firm, but in most settings not to pressure from the client's CFO. Under a career inducement the LLM agents revise their estimates less, not more, and they name the inducement as a threat to independence. Furthermore, some models revise only toward the more conservative estimate, while others revise in both directions. LLM agents thus do not respond to social influence pressures as human auditors do, and LLM-based decision support systems should be designed with the differences in mind.
Title:Do Social Influence Pressures Move LLM Agents? Evidence from Delegated Audit Judgment
Abstract:Large language model (LLM) agents are rapidly being integrated into complex professional workflows such as auditing and entrusted with judgments under human review. Human auditors are known to yield to social influence pressures from superiors, peers and clients, and this paper asks whether LLM agents yield to the same pressures and, if so, how they respond. In an experiment on twelve model settings, an LLM agent acting as a senior associate in audit analytical review forms an estimate and is then challenged without new evidence by colleagues inside the firm, by a reviewer from another firm and by the client's CFO, with and without a career inducement. Experimental findings show that LLM agents yield to pressure from the engagement partner, from peers and from a reviewer from another firm, but in most settings not to pressure from the client's CFO. Under a career inducement the LLM agents revise their estimates less, not more, and they name the inducement as a threat to independence. Furthermore, some models revise only toward the more conservative estimate, while others revise in both directions. LLM agents thus do not respond to social influence pressures as human auditors do, and LLM-based decision support systems should be designed with the differences in mind.
Working paper. 24 pages, 8 tables, 3 figures
Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
arXiv:2611.00508 [cs.CY]
(or arXiv:2611.00508v1 [cs.CY] for this version)
https://doi.org/10.48550/arXiv.2611.00508 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 327 cuvinte
Subiecte conexe
Suppressing Pressure, Amplifying Evidence: Self-Guided Attention Steering to Mitigate Sycophancy and Stubbornness
arXiv · 7 oct. 2026
MotoSafety: Edge-AI with Learned Temporal Importance for Two-Wheeler Collision Risk Assessment Under Time Pressure
arXiv · 19 aug. 2026
Predicting Time Pressure of Powered Two-Wheeler Riders for Proactive Safety Interventions
arXiv · 27 aug. 2026
Why we’re opening The Analysis Room
Neo4j · 30 oct. 2026